Mnist手写体识别实验报告

实验（二）

代码段：

# 实验环境：MindSpore-python3.7-aarch64

import os
# os.environ['DEVICE_ID'] = '0'

import mindspore as ms
import mindspore.context as context
import mindspore.dataset.transforms.c_transforms as C
import mindspore.dataset.vision.c_transforms as CV

from mindspore import nn
from mindspore.train import Model
from mindspore.train.callback import LossMonitor

context.set_context(mode=context.GRAPH_MODE, device_target='Ascend') # Ascend, CPU, GPU

data_train = os.path.join("MNIST", 'train') # train set
data_test = os.path.join("MNIST", 'test') # test set
ds = ms.dataset.MnistDataset(data_train)
print(data_train)

以上在华为云上不断报错，而且无法排除错误，于是我参考了网上视频教程，使用keras（基于tensorflow框架）第三方库在本地电脑上运行：

# 准备工作：

pip install keras
pip install tensorflow

第一次代码：

# 实验环境：pycharm, python 3.8
# 框架：tensorflow

from keras.utils import to_categorical
from keras import models, layers, regularizers
from keras.optimizers import RMSprop
from keras.datasets import mnist # 导入数据集
import matplotlib.pyplot as plt

# 加载数据集
(train_images, train_labels), (test_images, test_labels) = mnist.load_data() # 下载数据集
# print(train_images.shape, test_images.shape) # 打印输出形状
# print(train_images[0])
# print(train_labels[0])
# plt.imshow(train_images[0])
# plt.show()

# 将二维数据铺开成一维
train_images = train_images.reshape((60000, 28*28)).astype('float') # 784，输入层有784个神经元
test_images = test_images.reshape((10000, 28*28)).astype('float')
# 标签值编码
train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)
# print(train_labels[0])

# 搭建神经网络
# 输入层28*28个神经元，隐藏层15个神经元，输出层10个神经元（0~9）
network = models.Sequential() # 序列式模型
network.add(layers.Dense(units=15, activation='relu', input_shape=(28*28, ), )) # 隐藏层
# Dense：全连接层
# units：15个神经元
# 激活函数：ReLu（sigmoid和tanh会产生梯度弥散现象，自变量很大时，图像很平缓，梯度下降十分缓慢）
network.add(layers.Dense(units=10, activation='softmax')